With increasing operator experience, refinement in technology, and the availability of improved stent designs, precutaneous coronary intervention (PCI) is now considered the treatment of choice for many high-risk subgroups in which PCI was previously contraindicated. A number of PCI risk prediction/prognostic classification models have been described in the literature. These models are characterized by being inherently linear. Application of a more flexible and non-linear models for prognostic classification/risk prediction will potentially result into more interpretable prognostic classes with high clinical utility. Using the New York State Percutaneous Coronary Intervention Reporting System (PCIRS) database, this study proposes to develop a tree-structured prognostic classification for in-hospital mortality, post-procedural complications and long-term survival. In addition, the new tree-structured prediction models will be compared with the current prediction models described in the literature such as the Mayo Clinic Risk Adjustment model. Since tree-structured models are more intuitive, visual and based on a sequence of simple yes/no type questions involving clinical and patient characteristics that are familiar to the operating cardiologist they would maximize the clinical utility of prognostic classification. [unreadable] [unreadable] [unreadable]